Original Study

Journal of Veterinary Emergency and Critical Care 24(3) 2014, pp 272–278 doi: 10.1111/vec.12161

Measuring level of agreement between values obtained by directly measured blood pressure and ultrasonic Doppler flow detector in cats Anderson F. da Cunha, DVM, MS, DACVAA; Katrin Saile, DVM, MS, DACVS; Hugues Beaufr`ere, Dr. Med.Vet., PhD, DABVP (Avian), DECZM (Avian); Wendy Wolfson, DVM; Diana Seaton, BA, DVM and Mark J. Acierno, MBA, DVM, DACVIM

Abstract

Objective – To determine if blood pressure measured with an ultrasonic Doppler flow detector (Doppler) is in good agreement with directly measured blood pressures in anesthetized cats. Design – Prospective observational study. Setting – University veterinary teaching hospital. Animals – Thirty-nine cats undergoing routine neutering. Interventions – Cats were divided into 2 groups; 19 cats enrolled in Group A had a 24-Ga catheter inserted into a dorsal pedal artery; 20 cats in Group B had a 20-Ga catheter placed in a femoral artery. In both groups, systolic, diastolic, and mean arterial pressures were directly measured using a validated pressure measurement system. Indirect values were compared against direct blood pressure measurements. Results – There was no difference between groups. Overall, there was poor agreement with a significant bias observed between Doppler and directly measured blood pressures. For the systolic arterial pressure the bias was −8.8 with limits of agreements (LOA) of −39.3 and 21.7. For the mean arterial pressure, the bias was 14.0 with LOA of −13.9 and 41.9. For the diastolic arterial pressure, the bias was 27.9 with LOA of −4.4 and 60.2. Methodology, weight, sex, and replicates did not have a significant effect on the difference between indirect and direct measurements in any model. Conclusions – Results suggest poor agreement between Doppler values and directly measured blood pressures in anesthetized cats. Use of Doppler in cats could be misleading and readings should be interpreted with caution in a clinical context. (J Vet Emerg Crit Care 2014; 24(3): 272–278) doi: 10.1111/vec.12161 Keywords: blood pressure, cardiovascular, cat, Doppler, monitoring

Introduction Blood pressure measurement is essential for monitoring patients receiving anesthesia as well as for assessing critically ill patients and their response to therapy.1, 2 The clinical signs and significance of hypotension and From the Department of Veterinary Clinical Science, School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA 70810 (da Cunha, Saile, Wolfson); the Ontario Veterinary College, University of Guelph, ON, Canada (Beaufr`ere); and the Banfield Pet Hospital of Slidel, LA (Seaton). The authors declare no conflict of interests. Presented as an abstract at the International Veterinary Emergency and Critical Care Symposium/American College of Veterinary Anesthesia meeting, San Antonio, TX, September 2012. Address correspondence and reprint requests to Dr. da Cunha, Department of Veterinary Clinical Science, The Louisiana State University, School of Veterinary Medicine, Baton Rouge, LA 70803. Email: [email protected] Submitted September 20, 2012; Accepted January 01, 2014.

272

Abbreviations

AAMI ACVIM CI DAP Doppler LOA MAP SAP ␴2

American Association of Medical Instrumentation American College of Internal Medicine confidence interval diastolic arterial blood pressure Doppler flow detector limits of agreement mean arterial blood pressure systolic arterial blood pressure variance of a population

hypertension depend on the severity and duration of the underlying cause, as well as the severity and duration of the hypo- or hypertensive episode.3 Direct blood  C Veterinary Emergency and Critical Care Society 2014

Doppler measurement in cats

pressure measurement is considered the gold standard of blood pressure measurement; however, it is technically challenging, uncomfortable for the patient and unsuitable for many clinical situations.4–6 Several methods for indirect blood pressure monitoring are utilized in clinical practice; however, questions about the accuracy of each method have been raised.4, 6–10 The ultrasonic Doppler flow detector (Doppler) method has been widely used since 1977 in unrestrained cats.11, 12 It is generally considered inexpensive and easy to use. Nevertheless, questions remain as to its accuracy11, 13 and whether the reported values actually reflect systolic or mean arterial pressures.6, 14 In one study,6 blood pressure measurements obtained by Doppler were deemed to have low mean error values when compared with directly measured blood pressures; however, agreement between the 2 methods was poor. Results of another study7 indicated that although there was good correlation between directly measured systolic arterial blood pressures (SAP) and Doppler measurements, agreement analysis revealed a large negative bias (Doppler–SAP). Other authors8 have suggested that use of simple correction factors could improve the reliability of systolic blood pressures obtained by the Doppler. The purpose of this study was to determine if Doppler would measure systolic, diastolic or mean arterial pressures with good agreement with directly measured pressures in normal healthy cats. Our hypothesis was that Doppler readings would be in good agreement with the directly measured systolic blood pressure.

Methods and Materials The Louisiana State University clinical protocol and Institutional Animal Care and Use Committees approved the study protocol. Thirty-nine adult cats presented to the Louisiana State University Animal Sterilization Assistance Program for elective sterilization were enrolled in the study. The patients in this study lacked permanent homes and were presented by charitable organizations and animal shelters. All had incomplete medical histories but were deemed, on the basis of a physical examination, to be healthy for the sterilization procedure. In all cases, signed consent was obtained from the cat’s representative. Patients were sedated with IM midazolam (0.1 mg/kg), hydromorphone (0.1 mg/kg), and ketamine (7 mg/kg), and induced with isoflurane (3%) in 100% oxygen, administered by mask. All patients were intubated with an appropriately sized cuffed endotracheal tube and isoflurane anesthesia (1%) was administered through a Bain circuit. Partial pressure of carbon dioxide in the expired gas, partial pressure of isoflurane, and pulse oximetry were monitored with the use of a multi C Veterinary Emergency and Critical Care Society 2014, doi: 10.1111/vec.12161

function monitor.a Heart rate, esophageal temperature, respiratory rate, ECG, direct SAP, diastolic arterial blood pressures (DAP), and mean arterial blood pressures were recorded using a monitorb that digitally recorded the data. Cats were manually ventilated as needed to maintain the partial pressure of carbon dioxide in the expired gas between 35 and 45 mm Hg. A hot-water blanketc and a forced convective air warmerd were available to maintain body temperature between 98 and 100◦ F (36.5– 37.5◦ C), if needed. In order to evaluate if the agreement between Doppler and directly measured blood pressure was dependent on the accuracy of the pressure transducer, the location of the artery used and the catheter size applied, the cats were divided into 2 groups. In the first group (Group A; n = 19 cats) the direct blood pressure measurement was taken from a peripheral artery while in the second group (Group B; n = 20 cats), a more centrally located artery was utilized using different catheter diameters and pressure transducers. Blood pressure measurements for the 2 groups were performed as described below: Group A: The patients were placed in dorsal recumbency with the hind limbs extended caudally. After sterile preparation, a 24-Ga cathetere was placed in the dorsal pedal artery. The catheter was connected to a continuous multifunction monitor via a disposable pressure transducer systemf connected to a monitoring system.a The system was connected to a pressurized (300 mm Hg) bag of 0.9% saline. The transducer was placed at the level of the heart and zeroed to atmospheric pressure. The direct blood pressure monitoring system was visually inspected for air bubbles that could change the damping coefficient of the system and was periodically flushed to prevent clots. Direct pressure measurements were checked for stability, consistency and the waveform were analyzed before study recordings commenced. After the last measurement, arterial catheters were removed; pressure was applied to the dorsal pedal artery for 5 minutes. Group B: The patients were placed in dorsal recumbency, slightly tilted toward the operator, and the hind limbs were extended caudally. The femoral artery or arterial pulse was palpated and, after sterile preparation, a 2–3 cm skin incision was made parallel to the artery just distal to the femoral triangle. The femoral artery and vein were bluntly dissected from surrounding subcutaneous tissues and the femoral artery was isolated using 2–0 silk suturesg or a hemostat. A 20-Ga catheterh was placed in the femoral artery. The catheter was connected to a digital recording systemb via a high precision pressure transducer systemi using a disposable sterile dome connector.j This transducer’s 273

A. F. da Cunha et al.

output impedance is 1,000  and input impedance is 700  with a sensitivity of 5 ␮V/V/mm Hg. The direct blood pressure monitoring system was inspected for accuracy as described above. The transducer was placed at the level of the heart and zeroed to atmospheric pressure. After the last measurement, femoral catheters were removed; pressure was applied to the femoral artery for 5 minutes. Subcutaneous and intradermal tissues were sutured using 4–0 poliglecaprone 25.k Indirect blood pressure measurements were obtained with a Dopplerl using a human infant flat piezoelectric crystal probem (8.2 MHz). Hair over the palmar aspect of the carpus was clipped and conductive gel was applied to the Doppler probe which was secured directly perpendicular and over the digital artery with tape. A cuffn with a width of 30% to 40% of the circumference of the leg was utilized. The cuff was placed proximal to the transducer, in the middle third of the antebrachium. The cuff was inflated with a sphygmomanometero until sounds of pulsating blood generated by the unit were no longer audible. The cuff was gradually deflated until Korotkoff sounds were detected. Pressure displayed on the sphygmomanometer was recorded at that moment by a second individual (MA) who, at the same instant, noted the displayed arterial pressures from the multifunction monitor (group A) or added a note to the digital recording system in Group B. Afterward, in Group B, mean, systolic, and diastolic values were extracted from the digital recording system by calculating the average values 1 second before and after the annotation. The measurements were repeated 4 times at 1-minute intervals. All indirect blood pressure measurements were obtained by a single individual (AD) who was blinded to the direct measurements. Indirect blood pressure measurements were obtained from the front leg, ipsilateral to the arterial catheter. Both pressure transducers and sphygmomanometer were checked for accuracy against a mercury manometer using the 2 point method (50 and 150 mm Hg). No differences between sphygmomanometer, pressure transducers, and mercury manometer were observed. To validate the use of Doppler both the American Association of Medical Instrumentation (AAMI) standards and the American College of Internal Medicine (ACVIM) consensus statement15 on blood pressure measurement were used. The AAMI standards suggests that paired readings shall have a mean difference less than or equal to 5 mm Hg and a mean ± SD of less than 8 mm Hg.16 The ACVIM consensus statement15 on blood pressure measurement defines good agreement as a bias and limits of agreement within 15 mm Hg of the direct pressure. For the purpose of this study, good agreement was defined as a bias and limits of agreement within 15 mm Hg of the direct pressure as the ACVIM consensus recommends. 274

Statistical Methods Using a statistics software,p the agreement between the Doppler-derived and the direct arterial blood pressure was modeled using linear mixed modeling with the difference between indirect and direct measurements (Doppler value–direct value) as the response variable, groups (methodology), weight, sex, and replicates (repeated measures parameter) and blood pressure status hypotensive (MAP < 60 mm Hg), normotensive (MAP between 60 and 90 mm Hg), or hypertensive (MAP > 90 mm Hg) as fixed effects, and cats as random effect. The intercept of the model was equivalent to the bias (as determined by Bland and Altman17 ). The coefficient of reliability (consistency of measurements) was calculated by dividing the variance explained by the model (the cat random effect) by the total variance observed. The limits of agreement √ were obtained by computing the SE of the bias: 1.96 ␴ 2 with ␴ 2 the total variance of the model. Plots were performed as reported by Bland and Altman.17 Three different models were run for SAP, MAP, and DAP. For each model, SAP, MAP, and DAP were subsequently added, respectively, as fixed covariable effects to assess whether or not the difference of measurements (bias) depended on direct blood pressure values. Alternative models were also fitted where the Doppler measurements were modeled with the direct measurements as fixed effects and cats as random effect. The slope which corresponds to a proportional bias was tested against a slope of 1. Assumptions of the model (linearity, normality of residuals, and homoscedasticity of residuals) and influential data points were assessed by examining standardized residual and quantile plots. Autocorrelation of residuals over replicates was assessed by plotting the autocorrelation function and a likelihood ratio test between standard models and models with an autoregressive 1 covariance matrix. A positive bias (intercept of the model) indicated that the indirect Doppler measurement overestimated the direct measurement and a negative bias that it underestimated the direct measurement. A proportional bias 1 indicated that the indirect Doppler measurement overestimated the direct measurement with an overestimation that was proportional to the direct blood pressure.

Results Thirty-nine cats were included in the study. Cat weight was not normally distributed with a median of 3.4 kg and a range of 2.3 to 5.8 kg. There were 28 female and  C Veterinary Emergency and Critical Care Society 2014, doi: 10.1111/vec.12161

Doppler measurement in cats

Figure 1: Bland–Altman plot of agreement between direct systolic arterial blood pressure measurements and indirect measurements in 39 anesthetized cats undergoing routine neutering.

11 male cats. No age data could be collected due to the unknown origin of the animals. One cat was an outlier for SAP, MAP, and DAP with standardized residuals below −6 (below −2 to +3: potential outlier) and was removed from subsequent analysis. Methodology, weight, sex, and replicates did not have a significant effect on the difference between indirect and direct measurements in any model. A total of 624 measurements (4 times SAP, DAP, MAP, and Doppler in 39 cats) were obtained and used for statistical analysis. From all measurements, 25% were considered hypotension (MAP < 60 mm Hg), 54.2% were normotension (MAP between 60 and 90 mm Hg), and 20.8% of the measurements were considered hypertension (MAP > 90 mm Hg). However, the blood pressure state (hypotension, normotension, or hypertension) did not have a significant effect on the bias. For the SAP, there was a significant bias of −8.8 (95% CI: −12.0, −5.7, P < 0.001) with limits of agreements of −39.3 and 21.7. The coefficient of reliability was 0.65. This negative bias further decreased with increasing SAP values meaning that the technique was less reliable as SAP increases (coefficient of −0.13 [95% CI: −0.20, −0.06]) (Figure 1). The same effect was further confirmed by an alternative model in which a nonsignificant constant bias (intercept) with a proportional bias (slope) of 0.90 (95% CI: 0.88–0.91) was estimated. For the MAP, there was a significant bias of 14.0 (95% CI: 11.0–17.0, P < 0.001) with limits of agreements of −13.9 and 41.9. The coefficient of reliability was 0.66 (Figure 2).  C Veterinary Emergency and Critical Care Society 2014, doi: 10.1111/vec.12161

Figure 2: Bland–Altman plot of agreement between direct mean arterial blood pressure measurements and indirect measurements obtained with an ultrasonic Doppler in 39 anesthetized cats undergoing routine neutering. The dotted horizontal lines represent bias and 95% LOA as determined using linear mixed modeling.

Figure 3: Bland–Altman plot of agreement between direct diastolic arterial blood pressure measurements and indirect measurements in 39 anesthetized cats undergoing routine neutering.

For the DAP, there was a significant bias of 27.9 (95% CI: 24.3–31.4, P < 0.001) with limits of agreements of −4.4 and 60.2. The coefficient of reliability was 0.68 (Figure 3). All cats recovered well and were observed for a minimum of 5 days postoperatively. No hemorrhage, lameness, or other problems were reported after our procedure. 275

A. F. da Cunha et al.

Discussion The results of this study demonstrate that measurements made by the Doppler method were not in good agreement with directly measured pressures. A bias of −8.8 and 14.0 mm Hg was observed, respectively, for systolic and mean measurements when compared with Doppler readings. On average, all Doppler readings were 8.8 mm Hg lower than systolic and 14.0 mm Hg greater than mean. The large limit of agreement (LOA) (−39.3 to 21.7 mm Hg) suggests that indirectly measured systolic blood pressure values could be 39.3 mm Hg less or 21.7 mm Hg more than the directly measured values with 95% confidence. Furthermore, the wide limits of agreement suggest that Doppler is in poor agreement with direct systolic, mean, and diastolic blood pressure measurements and that no correction factor could reasonably be applied over the range of values obtained in our study. The large bias and wide LOA demonstrates that Doppler method is a poor estimator of either SAP or MAP in anesthetized cats. The interindividual variability cannot solely explain the wide LOA observed. Therefore, there have to be other not measured and identified factors influencing the variance (about a third of the measurement variability remains unexplained by factors accounted for in the model). Similar results have been observed in other previously published studies.9, 18 In a study on dogs, Doppler overestimated the systolic readings in the hypotensive group and underestimated systolic readings in the normotensive and hypertensive groups.18 Garofalo et al observed poor agreement of SAP determinations with the Doppler when SAP was higher than 140 mm Hg, regardless of cuff placement when comparing Doppler readings using 3 different cuff positions in anesthetized dogs.9 In the current study, although not significant, the agreement further deteriorated as SAP increased, whether they were normotensive or hypertensive. Previous studies6, 9, 19, 20 used pharmacologic manipulation to ensure that recorded blood pressures were distributed in low, normal, and high ranges. In the present study, this was not possible because all cats were clinical patients. Nevertheless, a wide range of blood pressures was obtained under standard clinical conditions. Because the Doppler unit was not able to produce measurements that were in good agreement with directly measured blood pressure in the normal ranges, there seemed to be little benefit of further exploring agreement for pressures in the hypertensive or hypotensive range, unless previously unmeasuredfactors that can influence the agreement are measured. Better agreement with direct measurement will likely depend on the development of higher precision indirect blood pressure measurement systems for use in animals of small sizes 276

instead of refining the Doppler method as used in this study. In this study, 2 different direct blood pressure measurement methods were applied. The first method used a transducer widely acceptable for the monitoring of the invasive blood pressure in clinical patients and a peripheral arterial catheter (dorsal pedal). For the second method, we used a highly accurate piezo-resistive transducer used for research purposes, and a larger bore catheter (20-Ga) applied at the femoral artery. We observed no statistical significant differences in mean measurements among methods. The size of the catheter, the artery catheterized, and the accuracy of the transducer were all variables suggested as possible factors that could add error to the blood pressure measurement however, in our study those factors did not appear to influence the final result, provided appropriate sample size was used. A previous study which evaluated the Doppler detector method of measuring systolic arterial blood pressure in cats suggested that adding a correction factor of 14.7 mm Hg to Doppler measurements would produce an accurate estimation of systolic pressure.8 Such a correction method would assume a constant variation between the Doppler and directly measured blood pressure. Examination of graphs 1, 2, and 3 demonstrates that the Doppler both over and under estimated directly measured blood pressures across a wide range of values. Furthermore, other variables such variability of the observer, the sphygmomanometer itself, the actual cuff and cuff placement and the number of readings would have to be constant in order to justify a simple correction factor for the data presented here. Therefore, the authors of this study consider a correction method not effective. Our data do not support that Doppler is an accurate means of measuring blood pressure in cats under anesthesia. However, during a single cuff placement in the same cat by the same observer, there is a possibility that the trend itself could be useful because most of the variability is expected to be due to cuff placement, individuals, and readings. Since our study did not investigate whether Doppler measurements could be used for trends during an anesthetic event with a single cuff placement in the same patient and measured by the same observer over time we cannot provide any conclusions on that aspect. Furthermore, based on our results, when assessing trends, the absolute values should not be regarded as valid. To determine whether Doppler would be useful to monitor trends of blood pressure, further studies would be needed to identify and quantify sources of variation (eg, cuff size, cuff placement, individuals, limbs of cats, observers) in Doppler readings in cats and whether they can be controlled during serial measurements in a single  C Veterinary Emergency and Critical Care Society 2014, doi: 10.1111/vec.12161

Doppler measurement in cats

anesthetic event keeping in mind the poor agreement with direct arterial blood pressure. There were some limitations to our study. Only one location was used for the Doppler placement and location of the cuff placement has been shown to alter the ability of the noninvasive blood pressure monitor to read correct arterial pressures in cats and other species.4, 21, 22 Additional studies should be performed to identify the most reliable anatomical location for the Doppler placement in cats. Additionally, tape was used to secure the Doppler crystal and in theory, different pressures could have been applied on the artery of different individuals explaining the interindividual variability; however, this was one of the authors’ concerns prior to the study and the authors did take time and care to apply the Doppler in a consistent manner to minimize chance of that possibility. The American Association of Medical Instrumentation (AAMI) has set standards for the performance of blood pressure devices which states that paired readings shall have a mean difference less than or equal to 5 mm Hg and a mean ± SD of

Measuring level of agreement between values obtained by directly measured blood pressure and ultrasonic Doppler flow detector in cats.

To determine if blood pressure measured with an ultrasonic Doppler flow detector (Doppler) is in good agreement with directly measured blood pressures...
419KB Sizes 0 Downloads 0 Views